Rizqy Hidayat

⚡ boring tech. 💼 code typer. 🌲 outdoor enthusiast.

[Rizqy Hidayat @rizqyhi]

Sebab di desa, waktu berjalan sedikit lebih lambat;
meski dunia tetap berputar sedemikian cepatnya.
Banyak hal berubah;
sebagiannya terpaksa mengikuti,
beberapanya tidak bisa.

[dex @dexhorthy]

Here’s what’s gonna happen:

- you replace your code review with feedback loops (sentry, datadog, support tickets, etc)
- you stop reading the code
- software factory fixes everything
- one day something breaks at 3am, agent can’t fix it
- nobody’s read the code in 3 months
- you have 3 weeks of downtime trying to re-onboard and fix it
- you lose significant % of your contracts and users
- your company is now dead

[Nav Toor @heynavtoor]

🚨BREAKING: OpenAI published a paper proving that ChatGPT will always make things up.

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

Every time ChatGPT gives you an answer, ask yourself: is this real, or is it just a confident guess?

[Alexey Grigorev @Al_Grigor]

Claude Code wiped our production database with a Terraform command.

It took down the DataTalksClub course platform and 2.5 years of submissions: homework, projects, and leaderboards.

Automated snapshots were gone too.

In the newsletter, I wrote the full timeline + what I changed so this doesn't happen again.

If you use Terraform (or let agents touch infra), this is a good story for you to read.

https://t.co/Mbi3oM4HMn

[ET Hadi Saputra 🍉🦉🇮🇩🏴‍☠️ @ethadisaputra]

Saya harus akui, jantung saya berdetak kencang, ingatan saya kembali menyergap tajam saat melihat berita bencana di Sumatera pekan ini.

Dulu, hampir dua dekade lalu, saya ada di sana. Bukan sebagai pengacara, melainkan penerbang sipil yang dipanggil negara, berdesakan di lambung pesawat tua. Ya, pesawat TNI AU uzur.
Kami terbang gila-gilaan, hampir tidak kenal pagi, siang, atau malam. Seringkali, kami sudah harus take off dari Halim sebelum subuh. Tujuannya: Banda Aceh.

Kami bawa Hercules C-130 yang sudah sepuh itu. Pesawat itu, Bapak-Ibu sekalian, adalah potret jujur bangsa ini: tua, berisik, kurang terawat, tapi untungnya sangat andal dan pekerja keras.
Di dalam perutnya yang gelap, kami duduk di jaring samping. Telinga berdengung raungan Allison T56.

Jika Anda mau tau, bau di dalam kabin itu adalah campuran unik: solar, minyak mesin yang bocor halus, keringat relawan yang tidak mandi tiga hari, dan aroma timpahan mi instan basi.

Intensitasnya? Luar biasa. Bandara Sultan Iskandar Muda (SIM) mendadak menjadi terminal paling sibuk di dunia. Ada Hercules TNI, ada C-17 Amerika, ada Ilyushin dari Rusia, semua berebut slot. Air Traffic Controller (ATC) kita seperti pesulap. Semua ingin cepat, semua jadi pahlawan tanpa berharap bintang.

Ironinya adalah pada sistem logistik kita.
Di satu sisi, kita punya Hercules TNI yang bertempur mati-matian, mengangkut puluhan ton barang. Di sisi lain, muncul pahlawan yang datang dari sektor yang paling tidak kita duga: Maskapai Sipil.
Ya, maskapai-maskapai besar seperti Lion Air dan Batavia Air ikut menyumbangkan free lift dari Jakarta. Mereka menggunakan pesawat komersial, kursi dilepas, diisi karung beras. Mereka bergerak karena kemanusiaan, mendahului banyak surat izin dan nota dinas yang mungkin masih diketik di kantor-kantor pusat.

Dan pahlawan sejati di udara adalah Susi Air. @susipudjiastuti Saat pesawat-pesawat besar TNI dan asing fokus ke Banda Aceh, Meulaboh dan Simeulue masih jadi titik buta. Landasan hancur, Susi dengan dua pesawat kecil Cessna Caravan-nya mengangkut obat-obatan dan susu bayi. Susi dan crew pilot asingnya terbang nekat, masuk ke landasan perintis. Mereka adalah antitesis dari birokrasi yang kaku. Contoh nyata bagaimana rule of law itu sejenak harus tunduk pada rule of need. Kebutuhan lebih dulu, baru administrasi.

Saya ingat, interaksi kami dengan crew Susi itu terasa sangat kontras. Di dekat Hercules yang penuh serdadu, mereka berdiri di samping Caravan kecil, mengenakan kaos, mengangkut sendiri kardus-kardus tanpa forklift mewah. Salah satu pilot asingnya pernah menyindir, "Saya pikir tugas saya hanya lobster, ternyata saya juga delivery harapan kemanusiaan." Tawa kami pecah. Tawa terdengar pahit tapi kejujurannya nyata hanya berharap pahala.

Saat ini, kita kembali menghadapi bencana di Sumatera. Apa yang berubah? Infrastruktur mungkin lebih baik. Teknologi komunikasi pasti lebih canggih. Namun, saya khawatir, jiwa gotong royong yang non-bureaucratic itu justru semakin menipis. yang jelas saya tidak dipanggil lagi menerbangkan pesawat, mungkin pilotnya sudah banyak. Mungkin juga karena memang tak ada landasan yang bisa didaratin fix wing. Mungkin juga kita makin terbiasa menunggu instruksi pusat, menunggu dana cair, menunggu SOP selesai dicetak.

Padahal, semangat yang dibutuhkan saat bencana adalah semangat Susi Air: bergerak cepat, tidak bertanya izin, dan langsung menuju titik yang paling terluka. Semangat yang harusnya diwarisi oleh setiap aparatur sipil dan militer.

Sudah saatnya kita belajar dari Hercules tua dan Cessna kecil.
Aset terbaik bangsa ini bukanlah pesawat baru atau regulasi yang tebal, melainkan keberanian mengambil risiko dan keikhlasan untuk bergerak tanpa menunggu tepuk tangan.

Jika tidak, setiap bencana hanya akan jadi pengulangan tragedi birokrasi yang mematikan.

#ethadisaputra #majalahforumkeadilan #tsunamiaceh #operasikemanusiaan #hercules #susiair #dahlaniskan #hukumdanlogistik #militersipil #bencanaindonesia

[Robert Graham @robertgraham]

So, CVE-2025-55182 is trending for good reason.

If you have external webapps in your company, your leaders need to force techies to come in on the weekend and fix it.

IT/security departments need to have an "emergency budget" when they force workers to stay late or come in on weekends to deal with an emergency, that they only do a couple times a year. This is one of them, the rest of your daily emergencies are not.

[Ryan Carniato @RyanCarniato]

This is pretty huge when you consider Nitro powers Nuxt, SolidStart, Tanstack Start, and AnalogJS, adding that to its existing funding of developers working on Next and SvelteKit.

Basically every metaframework except Astro, Remix, Qwikcity, and Fresh now has its core being developed by developers under @vercel. It's definitely something to think about.

[Rizqy Hidayat @rizqyhi]

finally a study that backed my takes on using AI.

[Alex Vacca @itsalexvacca]

BREAKING: MIT just completed the first brain scan study of ChatGPT users & the results are terrifying.

Turns out, AI isn't making us more productive. It's making us cognitively bankrupt.

Here's what 4 months of data revealed:

(hint: we've been measuring productivity all wrong)

[shadcn @shadcn]

Introducing the registry mcp. One command to make any component registry mcp-compatible.

Your Design System. Now with AI. Zero config.

We've got a lot to cover. Let's get started. ⬇️

[tobi lutke @tobi]

I heard this internal memo of mine is being leaked right now, so here it is:

[Adam Grant @AdamMGrant]

A neglected predictor of success and happiness is doing chores from an early age.

Kids who do regular chores in kindergarten get better math scores in third grade—and they're more satisfied with their lives.

Giving kids responsibility shows trust and builds character.

[Glauber Costa @glcst]

Here is one more Limbo story, the most mind-blowing so far.

After the announcement, we started receiving a large number of contributions from a man named Preston Thorpe (user PThorpe92).

All of them high quality contributions, implementing SQLite functions, fixing the io_uring event loop, getting us closer to supporting extensions, etc.

I came to learn that Preston is, in fact, in jail. He was jailed for non-violent drug offenses (I will link his story below), and is one of the first people in the United States to be granted permission to use the internet.

Most people in this situation, as he describes himself, are blemished for life, and will forever be going in and out of the system with no hope. But not Preston.

He went from a hopeless future to a top contributor on the open contribution evolution of SQLite (in rust!), and demonstrated for the whole world that he is capable of greatness.

Yesterday he sent me a message, with a screenshot of his name on the top contributors of the project and a red line over his name, saying
"I'll tell you what man, I don't think I've ever been so proud of something in my whole life"

Preston does not have access to X in prison, so likely he will not read this. But if you do: no, man. *We* are proud of you.

We are proud of the example that you are setting, the path that you are offering to other people in your situation, and you are a part of the reason we have decided to change our entire company strategy to make sure we have the resources needed to make sure this project succeeds and is ready to replace SQLite in production ASAP.

I'll leave his links below 👇

[serafim @korablev]

We’re live on Product Hunt!

• 800+ components from 50+ top design engineers
• Add with copy & paste or @shadcn CLI—code, dependencies, hooks, CSS, and Tailwind config included
• Optimized for AI editors like @lovable, @stackblitz, bolt, and @vercel’s @v0

👉 Link in the next post—please support!

[Ibu Kancil @wulankuswo1]

@kozirama Kebersyukuran tuh levelnya individu. Kebijakan yang meningkatkan kualitas hidup manusia tuh levelnya pemerintah. Jangan suruh masyarakat bersyukur utk kebijakan yg buruk. Tuntut agar lebih baik.

[Okky Madasari @okkymadasari]

Diberi makan siang gratis, tapi dirampas haknya untuk berpikir, menilai, menyampaikan pendapat. Terbungkam dan hidup dalam ketakutan?

[Obbe Vermeij @ObbeVermeij]

20 years ago we released Grand Theft Auto - San Andreas.

The original plan was for the 3 cities to be on separate maps. The player would travel between the cities using trains and planes.
(Gta 1 and 2 also had three cities on separate maps)

Memory was very tight on the ps2 and with separate maps, the other cities’ skyline models wouldn’t have to be in memory.

It would also make it easier to have different police/ambulance/firetrucks for each city. Different pickups, weather types etc.

It would also be easier to contain the player until it was time for the next city.

It would also make it easier to organize the models on the DVD city-by-city which would help the streaming.

Just before the artists started working on the three maps, we had a final meeting at R* North in which we changed our minds and decided to go for a big map after all.

We still ended up doing city-specific pickups, police cars and weather.

Happy birthday San Andreas. You turned out all right.

[Rizqy Hidayat @rizqyhi]

Currently tryin WordPress' Site Editor with Twenty Twentyfour theme and I'm officially lost. It's totally different from Customizer era. Might not that bad, but I'm just not ready for it.

[Daniel Lockyer @DanielLockyer]

Vercel pricing is absolutely insane, it blows my mind

Data transfer starts at $150/TB with 1TB included

Hetzner gives 20TB free, and then it's €1.19/TB

So 20TB on Vercel is $2850, on Hetzner it's free (included with VPS)

And then you have to pay extra per request?

Perhaps apples and oranges but I haven't even included half the other things you need to run apps on Vercel

[Marko Denic @denicmarko]

Learn programming by playing games:

Blockchain → cryptozombies.​io
SQL → mystery.knightlab.​com
JavaScript → warriorjs.​com
Go → codingame.​com
Python → codedex.​io
CSS → cssbattle.​dev
Java → tynker.​com
Git → ohmygit.​org

[Rizqy Hidayat @rizqyhi]

Can I get the swipe-from-bottom gesture back? :(

[Motasem A Dalloul @AbujomaaGaza]

My dear followers,

I need your support to reach 100,000 supporters of the truth and the right of Palestinians..

Mention my account, repost my posts, refer my account to friends, etc..

Do whatever you can to help me reach this target..